Summary of Disents: Disentangled Channel Evolving Pattern Modeling For Multivariate Time Series Forecasting, by Zhiding Liu et al.
DisenTS: Disentangled Channel Evolving Pattern Modeling for Multivariate Time Series Forecasting
by Zhiding Liu, Jiqian Yang, Qingyang Mao, Yuze Zhao, Mingyue Cheng, Zhi Li, Qi Liu, Enhong Chen
First submitted to arxiv on: 30 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed DisenTS framework addresses the limitations of mainstream multivariate time series forecasting methods by modeling disentangled channel evolving patterns. The framework employs multiple distinct forecasting models, each tasked with uncovering a unique evolving pattern. A novel Forecaster Aware Gate (FAG) module is introduced to generate routing signals adaptively according to both forecasters’ states and input series characteristics. Additionally, the Similarity Constraint (SC) is proposed to guide each model to specialize in an underlying pattern by minimizing mutual information between representations. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary DisenTS helps predict future events in many real-world situations. Currently, most methods use one single model that tries to understand all the different channels (like temperatures or stock prices) together. But sometimes, these channels can have very different patterns and trends. DisenTS is a new way of modeling this data by using multiple smaller models, each trying to understand its own unique pattern. It also has special gates that help figure out which model should be used for a particular forecast. |
Keywords
» Artificial intelligence » Time series
